Thread: Stale Process
Dear sir,
I built Apache, PHP-PostgreSql 7.1 web server in Linux
Redhat 7.0.
I always check the process by 'ps -aux' and always
found the stale postmaster processes:
postgres 9212 0.0 2.1 6136 2408 ? S
Nov09 0:02 postgres: nobody newmy3qdb [local] idle
postgres 10617 0.0 2.4 5628 2740 ? S
Nov09 0:02 postgres: nobody newmy3qdb [local] idle
The processes would not disappear if I don't kill the
processess.
I only use 'pg_connect' but never use
'pg_pconnect'(permanent connection), so why does it
happen and how to solve the problem so that the stale
process can automatically disappear?
=====
Yours sincerely,
Leung Chun Man, Terence
Mobile: (852) 9273 9176
Homepage: http://tcmleung.uhome.net ,
http://www.dynamicdrive.com/dynamicindex4/filter/index.htm
__________________________________________________
Do you Yahoo!?
U2 on LAUNCH - Exclusive greatest hits videos
http://launch.yahoo.com/u2
Terence Leung <tcmleung@yahoo.com> writes: > Dear sir, > I built Apache, PHP-PostgreSql 7.1 web server in Linux > Redhat 7.0. > I always check the process by 'ps -aux' and always > found the stale postmaster processes: It's most likely that PHP is keeping the connections open for some reason. The 'postgres' processes are in S (sleep) state, which indicates that their socket is still being held open by the client end. So it's probably not a Postgres problem... I find it entirely unsurprising that PHP might be buggy. ;) -Doug
On 10 Nov 2002, Doug McNaught wrote: > Terence Leung <tcmleung@yahoo.com> writes: > > > Dear sir, > > I built Apache, PHP-PostgreSql 7.1 web server in Linux > > Redhat 7.0. > > I always check the process by 'ps -aux' and always > > found the stale postmaster processes: > > It's most likely that PHP is keeping the connections open for some > reason. The 'postgres' processes are in S (sleep) state, which > indicates that their socket is still being held open by the client > end. So it's probably not a Postgres problem... I find it entirely > unsurprising that PHP might be buggy. ;) Yes, he should switch immediately to a bug free language, like maybe Java. It's a poor craftsman who blames his tools. Anyway, now that we've gotten our jabs in on our non-favorite language, mayhaps we can help the newbie? I've never seen PHP hold a connection open when it wasn't opened using pconnect. It's possible that some page on the box DOES have a pg_pconnect in it somewhere. That's where I'd look first. It's also possible that a user has a psql connection up and running. If you've got a psql monitor open, you'll see an idle backend process waiting to take you queries and run them. Lastly, a single idle process is NO BIG DEAL. If your database only has two connections to it at idle, and 100 at heavy load, the extra idle connection is below the noise threshold of "things to be worried about."
On Sun, 10 Nov 2002, Terence Leung wrote: > Dear sir, > I built Apache, PHP-PostgreSql 7.1 web server in Linux > Redhat 7.0. > I always check the process by 'ps -aux' and always > found the stale postmaster processes: > > postgres 9212 0.0 2.1 6136 2408 ? S > Nov09 0:02 postgres: nobody newmy3qdb [local] idle > postgres 10617 0.0 2.4 5628 2740 ? S > Nov09 0:02 postgres: nobody newmy3qdb [local] idle > > The processes would not disappear if I don't kill the > processess. > I only use 'pg_connect' but never use > 'pg_pconnect'(permanent connection), so why does it > happen and how to solve the problem so that the stale > process can automatically disappear? As a followup to my previous message, the other possibility is that you have a PHP page that is segfaulting halfway through, and the connection is being left hanging since the process died before it could close it. But the other two possibilities are more still more likely (i.e. psql is running, or you have a pg_pconnect somewhere you haven't found yet.)